the effect of overseas investment on domestic employment · the majority of overseas investment,...

This PDF is a selection from a published volume from theNational Bureau of Economic Research

Volume Title: International Trade in East Asia, NBER-EastAsia Seminar on Economics, Volume 14

Volume Author/Editor: Takatoshi Ito and Andrew K. Rose,editors

Volume Publisher: University of Chicago Press

Volume ISBN: 0-226-37896-9

Volume URL: http://www.nber.org/books/ito_05-1

Conference Date: September 5-7, 2003

Publication Date: August 2005

Title: The Effect of Overseas Investment on Domestic Employment

Author: Tain-Jy Chen, Ying-hua Ku

URL: http://www.nber.org/chapters/c0192

109

4.1 Introduction

It has long been a concern of policymakers that foreign direct invest-ment (FDI) may cause job losses at home; indeed, labor unions generallyconsider FDI to be the equivalent of job exporting. The logic is simple; asproduction lines are relocated overseas, gone with them are the workersthat served the domestic lines. This reasoning is, of course, oversimplisticbecause there could never be any guarantees that the production lines thatwere relocated overseas would have been able to survive the competitionhad they remained at home. If these production lines were to be eliminatedanyway, then their relocation does not result in any job losses.

Conversely, there is always the possibility that overseas investment mightwell enhance the overall competitiveness of the investing company andtherefore boost job opportunities at home that would otherwise have beenswept away by competition. Ku (1998), for example, found that FDI en-abled Taiwanese enterprises to restructure themselves and therefore in-crease their tenacity. She showed that firms engaging in overseas produc-tion had a better chance of survival than those that were not.

Those who are concerned about the adverse effects of overseas invest-ment on domestic employment basically assume that overseas productionis a substitute for exports; hence, as exports fall, so does employment. Thisis a conventional argument along the lines of Mundell (1957), who showedvery elegantly, in a 2 � 2 model, that capital movement is equivalent totrade. Products produced in overseas locations not only replace exports,

4The Effects of Overseas Investmenton Domestic Employment

Tain-Jy Chen and Ying-Hua Ku

Tain-Jy Chen is president of the Chung-Hua Institution for Economic Research and a pro-fessor in the department of economics at National Taiwan University. Ying-Hua Ku is a re-search fellow at the Chung-Hua Institution for Economic Research.

they may also in fact be reimported back home to substitute the productsthat were previously produced to serve the home market (Liu and Lin2001). There are, however, counterarguments to Mundell’s perfect substi-tution theory. Markusen (1983), for example, demonstrated the theoreticalpossibility that FDI and trade are complementary rather than substitutes;therefore, the relationship between FDI and job opportunities at home isindeed an empirical question.

Brainard and Riker (1997a,b) directly estimated the substitution elastic-ities between employment in parent companies and their foreign affiliates,as well as those between different affiliates, and found a very low degree ofsubstitution between parent and affiliate employment, although there wasa high degree of substitution between affiliates in developing countries.They also found that the relationship between employment in industrial-ized country affiliates and in developing countries was complementaryrather than substituting.

Slaughter (1995) had earlier found a similar low degree of substitutionbetween parent and affiliate employment when only production workerswere considered. He noted that the employment of production workers didnot seem to be systematically related to relative wages between the parentand the affiliate. This suggests that overseas employment corresponds onlyweakly to the wage gap between home and host countries, although it maycorrespond strongly to the wage gap between different overseas locations.Hatzius (1997) and Döhrn (1997) found similar results for Sweden withoverseas employment of Swedish multinational firms responding to wagesin actual and potential host countries but not to wages in Sweden. Blom-ström and Kokko (2000) also discovered that Swedish multinationals reactto domestic policies rather than wages in determining whether to keep pro-duction at home.

This evidence suggests that overseas production and domestic produc-tion is closely related but not necessarily substitutable. In fact, there mustbe a division of labor between the parent and affiliates as FDI is an actiontaken to enhance the competitiveness of a company. To the extent that FDIreduces the costs of the parent’s operations, it also helps the parent to ex-pand its level of output, which, in turn, increases employment at home.Blomström, Fors, and Lipsey (1997), for example, found that overseas in-vestment in developing countries by U.S. firms did have the effect of replac-ing domestic employment, but the same investment in developed coun-tries did not; the replacement effect was, however, limited to productionworkers.

Findings that the employment effect from FDI may differ across laborgroups are important, for this implies that FDI has important conse-quences for income distribution. For example, the examination of Swedishfirms by Blomström, Fors, and Lipsey (1997) found that FDI contributesto growth in employment of unskilled labor at home because Swedish

110 Tain-Jy Chen and Ying-Hua Ku

multinationals were investing abroad to acquire skilled workers to engagein R&D and other skill-intensive activities. Lipsey’s (1994) study of U.S.multinationals also found that overseas affiliates allow the parent to em-ploy more managerial and technical staff at the same level of domesticproduction. Feenstra (1996) showed that FDI in Mexico by U.S. firmsincreased the demand at home for skilled workers vis-à-vis unskilledworkers, thus raising the relative wage of skilled workers and worsening in-come distribution for the investing country, whereas the reverse occurredin Mexico.

There is an indirect, but nevertheless very important, linkage betweenFDI and domestic employment, that is, the effect of FDI on domestic in-vestment. If FDI outflows are accompanied by an equal reduction in theamount of domestic investment, then FDI may still reduce job opportuni-ties at home even if overseas production is complementary to domesticproduction; Feldstein (1994) seems to suggest such a one-to-one substitu-tion effect. Stevens and Lipsey (1992) also found a negative relationship be-tween FDI and domestic investment, although not as clear as one-to-onereplacement, however, Bayoumi and Lipworth’s (1998) study of the case inJapan found no displacement effect on domestic investment from FDI.Again, the actual relationship is therefore an empirical question.

The purpose of this paper is therefore to examine the relationship be-tween FDI and domestic employment at firm level, using Taiwan’s manu-facturing industry as an example. We find that overseas production leadsto an increase in the domestic employment of managerial and technicalworkers but may also reduce the employment available to unskilled work-ers. Overseas production partially replaces inputs to domestic production,resulting in a decline in labor demand at a given output level; however, atthe same time, overseas production reduces the costs of domestic produc-tion, leading to an expansion in output. These input-replacement and out-put-expansion effects combine to produce a net effect that is positive inmost cases, although the net effect differs with different labor groups andthe geographical location of overseas investment.

4.2 An Overview of Taiwan’s FDI and Manufacturing Employment

Taiwanese firms made only sporadic outward investment before 1980.Beginning in the mid-1980s, Taiwanese firms started making more sub-stantial foreign investment, driven by rising wages and rising value ofTaiwanese currency, NT. Between 1987 and 1990, Southeast Asia and theUnited States were the major destinations of Taiwan’s foreign investment.In the early 1990s, China emerged in the FDI map and eventually becamethe most popular destination for Taiwanese investors. In the second half ofthe 1990s, China took up almost a half of Taiwan’s total amount of out-ward investment (see table 4.1). The manufacturing sector accounted for

The Effects of Overseas Investment on Domestic Employment 111

the majority of overseas investment, dominating the service and agricul-ture sectors. In the manufacturing sector, FDI is most active in the elec-tronics, chemical, and textile industries. FDI appears to have importantconsequences on domestic employment.

Manufacturing employment in Taiwan reached a peak in 1987 when2.821 million people were working in the manufacturing sector; thereafter,there was a general decline in manufacturing employment until it hit atrough in 1994, when 2.422 million people were working in the sector. Itthen started to recover through the mid- to late-1990s, with 2.655 millionpeople being employed in the manufacturing sector by 2000 (see figure4.1).

The available employment data suggests that the period 1987–1994 wasa time when Taiwan’s industry underwent dramatic restructuring. Whilethere were losses of 399,000 manufacturing jobs throughout that period,there was nevertheless an increase in employment in the service sector ofaround 1.385 million, more than enough to offset these losses. Thus, un-employment rates remained at low levels throughout the 1990s.

It is also worth noting that 1987 was around the time when Taiwanesefirms began to embark on the course of FDI, with more than US$43 bil-lion being invested overseas from 1987 to 2000. Between 1987 and 1992,FDI was concentrated in Southeast Asia where Malaysia, Thailand, andIndonesia took the lion’s share of Taiwan’s overseas investment; however,from 1992 onward, the focus for FDI shifted to China. After the 1997Asian financial crisis, FDI in Southeast Asia came to a virtual standstill


Table 4.1 Taiwan’s outward investment by location (US$1,000)

Asia (excluding China) America Europe China Others Total

1952–1990 1,077,710 1,844,332 115,171 0 39,298 3,076,5111991 929,819 658,958 60,289 174,158 6,964 1,830,1881992 369,929 449,096 45,933 246,992 22,301 1,134,2511993 663,514 740,110 255,913 3,168,411 1,398 4,829,3461994 559,471 988,336 22,209 962,209 46,748 2,578,9731995 467,743 787,105 59,868 1,092,713 42,162 2,449,5911996 661,717 1,442,953 11,875 1,229,241 48,859 3,394,6451997 818,743 1,915,948 58,508 4,334,313 100,627 7,228,1391998 580,819 2,637,021 33,828 2,034,621 44,634 5,330,9231999 836,378 2,267,710 60,982 1,252,780 103,943 4,521,7932000 851,065 3,946,021 62,225 2,607,142 217,751 7,684,2042001 814,981 3,460,902 45,594 2,784,147 70,177 7,175,8012002 528,054 2,475,575 123,416 6,723,058 243,001 10,093,104

Total 9,159,943 23,614,067 955,811 26,609,785 987,863 61,327,469

Source: Statistics on Overseas Chinese & Foreign Investment, Outward Investment, and Indirect MainlandInvestment (various issues), Investment Commission, Ministry of Economic Affairs.

whereas FDI in China continued to surge. In 2001, the global recession sawTaiwan’s unemployment rate reaching an unprecedented 4 percent; thus,there were heightened fears that FDI may have led to rising unemploymentat home.

Beneath the surface of a relatively stable employment situation in the1990s, there was a rather dramatic transformation taking place in the in-dustrial structure. Among twenty-two two-digit industries in the manufac-turing sector, twelve had increased their employment levels whereas theremaining ten had seen their employment levels falling. The most rapidincrease in employment occurred in the electronics industry in which145,748 new jobs had been generated between 1991 and 2000, representinga 24.3 percent increase on the 1990 level. It was probably no coincidencethat the electronics industry was also the industry that was most active inundertaking outward investment. In contrast, employment in the apparelindustry recorded the largest number of job losses, at 54,104, representinga loss of more than one-third of its initial 1991 employment level. However,FDI from the apparel industry was also substantial; thus, the relationshipbetween FDI and domestic employment is unclear, to say the least. In thefollowing section, we will examine this relationship in more detail.

4.3 The Statistics on FDI and Employment

In this section, we present the employment data revealed by Taiwan’sCensus of Manufacturers and relate this to FDI. The census data are col-lected at plant level but are then aggregated into firm-level data; all of the


Fig. 4.1 Manufacturing employment, 1981–2000

following statistics are reported at firm level because it is considered thatFDI is decided at firm level rather than at plant level. Changes in employ-ment between 1993 and 2000 are studied, with 1993 having been chosen asthe starting year because this was the first time that a comprehensive set ofFDI statistics was collected in the census; 2000 is chosen as the terminalyear because this was the most recent census year. A total of 75,101 firmsare included in the 1993 census, of which 49,260 had survived until 2000,while the remaining 25,841 had exited the market during the period understudy. Between 1993 and 2000, 27,585 new firms had entered the market,with these new entries during this eight-year period representing 36.7 per-cent of the stock of firms in the initial year, and the exiting firms repre-senting 34.4 percent of the stock, a characteristically high turnover rate forTaiwan’s industry (Aw, Chen, and Roberts 2001). All firms that have shownup in either the 1993 census or the 2000 census come to a total of 102,686,which forms our sample for comparison.

We classify all sample firms into two categories, the FDI group and thenon-FDI group. The FDI group includes all firms that have undertakenoverseas investment, and the non-FDI group includes those that have notundertaken any such investment. Although there are some missing data,the census does cover the majority of manufacturing firms. The total em-ployment figures in the sample were 2,155,672 persons for 1993, and2,291,396 for 2000, representing 89.8 percent and 92.9 percent, respec-tively, of the total employment estimated by the statistics authorities dur-ing the two census years.

We tabulate the turnover of sample firms in table 4.2, which shows thatthere were 4,283 firms in the FDI group and 98,403 firms in the non-FDIgroup. Although, in terms of the number of firms, the FDI group ac-counted for just 4.3 percent of the manufacturing sector (ignoring the


Table 4.2 FDI and domestic employment, 1993–2000 (persons, %)

1993 2000

Firm group No. of firms Employment % Employment %

FDI firms 4,283 608,501 28.23 689,769 30.10Survivors 2,843 558,243 25.90 625,013 27.28Exited 900 50,258 2.33 n.a. n.a.New entrants 540 n.a. n.a. 64,756 2.83

Non-FDI firms 98,403 1,547,171 71.77 1,601,627 69.90Survivors 46,417 1,119,060 51.91 1,055,421 46.06Exited 24,941 428,111 19.86 n.a. n.a.New entrants 27,045 n.a. n.a. 546,206 23.84

Total 102,686 2,155,672 100.00 2,291,396 100.00

Source: Authors’ calculation from Census of Manufacturers, 1993 and 2000.Note: n.a. � not available.

missing data), it nevertheless accounted for 28.23 percent of total employ-ment within the sector, which suggests that firms engaging in overseas in-vestment are relatively large in size.

Out of the 4,283 firms in the FDI group, 3,743 firms were already in ex-istence in 1993; the remainder was made up of new firms that entered dur-ing the period under study. From the initial 1993 cohort, 2,843 had sur-vived the competition and remained active within the industry in 2000,representing a 76.0 percent survival rate.

Meanwhile, out of the 98,403 firms in the non-FDI group, 71,358 firmswere already in existence in 1993, and 46,417 firms had survived up until2000, representing a 65.0 percent survival rate. Simple statistics suggestthat those firms that were engaged in overseas investment had a higher sur-vival rate, supporting the findings of Ku (1998), which, in a study of Tai-wan’s electronics industry, showed that FDI did indeed increase the prob-ability of survival.

Within our sample, the FDI group accounted for 28.23 percent of allmanufacturing sector employment in 1993, but by 2000, this figure hadrisen to 30.10 percent. If we count only those firms that were in existencein 1993, the employment share in 2000 was 27.28 percent, representingonly a slight fall on the 1993 proportion despite the fact that a quarter ofthem had been eliminated in the interim period. In contrast, the non-FDIgroup accounted for 71.77 percent of all manufacturing sector employ-ment in 1993 and 69.90 percent in 2000. However, if new entrants are ex-cluded, the surviving firms in the non-FDI group account for only 46.06percent of employment in 2000. Simple statistics again suggest that FDIenabled investing firms to maintain more jobs at home.

It is worth noting that firms that exited the manufacturing industryduring the period under study eliminated 478,369 jobs, or 22.2 percent ofthe total employment in 1993. These losses were more than offset by the610,962 jobs created by new entrants coming into the industry during theeight-year period. Total employment provided by those firms that survivedthe period is virtually unchanged; however, employment per firm increasedby 12.0 percent in the FDI group in contrast to the 5.7 percent decline inthe non-FDI group.

4.4 The Effect of Investment Location

As demonstrated by Lipsey (1994) and Blomström, Fors, and Lipsey(1997), the employment effect of FDI may differ by investment location;thus, we should also examine the data on Taiwan to see whether geograph-ical location matters. Taiwanese FDI had been concentrated in China sincethe early 1990s; however, there is one perspective that argues that invest-ment in China is potentially more harmful to domestic employment thanFDI in other regions. The reason for this, so the argument goes, is because


of the cultural proximity and similarity in labor skills, with production inChina being likely to duplicate what had previously been done in Taiwanand therefore exerting a strong substitution effect on domestic employ-ment.

In order to examine the location effect, we classify those firms under-taking overseas investment into four subgroups according to the locationof their investment. The first subgroup contains firms undertaking invest-ment in China only, the second subgroup contains firms investing in Chinaplus other regions, the third subgroup contains firms investing in regionsother than China, and the fourth subgroup contains firms with unknownFDI locations. Table 4.3 provides details of the level of employment for thefour respective subgroups in 1993 and 2000.

As the table shows, of the 2,843 firms that undertook overseas invest-ment and survived the 1993–2000 period, 1,048 had invested only in China,630 had invested in China and somewhere else, 692 had invested only out-side of China, and the remainder had invested in unknown regions. Thoseinvesting only in China were apparently smaller in size as their average em-ployment was only 116.58 in 1993, substantially lower than the average employment level for the entire FDI group; furthermore, the average em-ployment of this subgroup declined again, to 107.55 employees, in 2000. Incontrast, the subgroup investing only outside of China registered the high-est growth rate in employment of all the subgroups, at 33.49 percent, whilefirms that invested in China and other regions saw their employment riseby 16.99 percent.

This seems to suggest that investing only in China undermines the in-vestor’s capacity to maintain jobs at home; however, this conclusion issomewhat premature as there are other factors that may affect domesticemployment after an enterprise invests abroad. Two obvious factors arefirm size and industry. It is well established within the literature that firmsize is positively correlated to the ability to invest abroad (Caves 1971,1996). Large firms may therefore be more capable of undergoing internalrestructuring after they have invested abroad and therefore are more ca-pable of maintaining jobs at home (Chen and Ku 2000).


Table 4.3 Employment effect, by FDI location

Investment No. of Average Average 1993–2000 location firms 1993 employment 2000 employment change (%)

China 1,048 122,179 116.58 112,710 107.55 –7.75China and others 630 284,876 452.18 333,269 529.00 16.99Other than China 692 101,698 146.96 135,752 196.17 33.49Unknown 473 49,490 104.63 43,282 91.51 –12.54

Total 2,843 558,243 196.36 625,013 219.84 11.96

Source: Authors’ calculation from Census of Manufacturers, 1993 and 2000.

Industry is also an important factor because a high-growth industry pro-vides more opportunities for firms to diversify after they have investedabroad. In order to test the size and industry effects, we make a two-wayclassification of firms according to their size and industry affiliations; firmsthat employ more than 300 persons are classified as large firms; the rest aresmall firms. Industries that have grown by more than 30 percent in outputbetween 1993 and 2000 are considered to be high-growth industries; oth-erwise they are low-growth industries; the demarcation line of 30 percentis the average growth rate in entire manufacturing output for the period un-der study.

We apply analysis of variance (ANOVA) to determine how much FDIlocation matters when controlling for industry and size and vice versa; theresults are shown in table 4.4, which indicates that when controlling for in-vestment location, employment growth is significantly affected by both in-dustry and size. Firms in the high-growth industries show a significantlyhigher employment growth rate than those in the low-growth industries,while large firms show a significantly higher employment growth rate thansmall firms.

When both industry and firm size are controlled for, investment locationbecomes inconsequential, except for the small-firm group where those in-vesting in China only registered the lowest employment growth rate, ascompared to those investing outside of China. This seems to suggest thatjob displacement, if there is any, may affect small firms that choose to in-vest solely in China.

4.5 Estimating the Effects of FDI on Employment

In this section, we estimate the statistical effects of FDI on employment,using a production function to portray the relationship between domesticand overseas operations. We basically treat overseas operation and do-mestic operation as joint production that can be portrayed by an appro-


Table 4.4 Change in employment, 1993–2000 (ANOVA)

Industry Size

Investment location Low growth High growth F-statistics Small Large F-statistics Sample

China 0.017 0.812 3.27∗ 0.044 4.238 27.43∗∗ 1,048China and others 0.235 1.524 8.52∗∗ 0.628 1.669 27.15∗∗ 692Other than China 0.183 0.532 4.59∗∗ 0.241 1.239 2.92∗ 630Unknown 0.226 0.716 0.458 0.525 473

F-statistics 1.36 1.01 2.61∗∗ 1.19 2,843

∗∗Significant at the 5 percent level∗Significant at the 10 percent level.

priate production function. The output from overseas production mayserve as an intermediate input to domestic production, thereby reducingthe cost of domestic production; by so doing, this reduces the demand fordomestic primary inputs, including labor. The output from overseas pro-duction may also add to the burden of domestic operations if it requiresmanagerial and technical support from the headquarters. Here, we treatthe output from both overseas and domestic operations as two jointoutputs from centrally managed production aimed at minimizing overallcosts.

We employ the generalized Leontief production function developed byDiewert (1971) and Hall (1973) to portray a cross-border operation yield-ing two distinctive outputs Y1 and Y2, where Y1 is the output from domes-tic operations and Y2 is that from foreign operations. There are three kindsof labor inputs to production, namely managerial workers, technical work-ers, and blue-collar workers. Labor is finely classified because we are con-cerned about the effects of FDI on different kinds of labor, given thecomplexity of the international division of labor. Three kinds of workersconstitute a composite labor input underlying which is a subproductionfunction. The relationship between this composite labor input and capitalis a Leontief relationship; therefore, the demand for labor can be solely de-termined by output levels and wages, irrespective of capital input. We cantherefore depict the cost function of the composite labor as follows:

C (Y1, Y2, W1, W2, W3 ) � �1Y1W1 � �2Y1W2 � �3Y1W3 � �4Y2W1 � �5Y2W 2

� �6Y2W3 � 2�7W1�Y1Y2� � 2�8W2�Y1Y2�

� 2�9W3�Y1Y2� � 2�10Y1�W1W2� � 2�11Y1�W1W3�

� 2�12Y1�W2W3� � 2�13Y2�W1W2�

� 2�14Y2�W2W3� � 2�15Y2�W1W3�

� 4�16�Y1Y2W�1W2� � 4�17�Y1Y2W�2W3�

� 4�18�Y1Y2W�1W3�,

where C is the total cost of labor and W1, W2, and W3 are the respective unitcosts of managerial workers, technical workers, and blue-collar workers.Note that outputs Y1, Y2 are measured by value added in NT dollar terms.The sample covers firms from various industries, and value added is theonly meaningful measuring unit common to all industries.

Although the generalized Leontief production function restricts the pro-duction technology to be constant returns to scale, it does allow the elas-ticity of substitution (or complementarity) between three kinds of labor tobe flexible. The interrelationship between different kinds of labor in pro-duction is the focus of our study.

Using Shephard’s lemma, we may derive the labor demand equation foreach kind of worker:


(1) L1 � �∂∂W

C

1

� � �1Y1 � �4Y2 � 2�7�Y1Y2� � �10Y1��W

W2

1

�� 11Y1��W

W3

1

�� 13Y2��

W

W2

1

�� 15Y2��W

W3

1

�� 2�16��Y1Y

W2

1

W�2��

� 2�18��Y1Y

W2

1

W�3��

L2 � �∂∂W

C

2

� � �2Y1 � �5Y2 � 2�8�Y1Y2� � �10Y1��W

W1

2

�� 12Y1��W

W3

2

�� 13Y2��

W

W2

1�� 14Y2��

W

W3

2

�� 2�16��Y1Y

W2

2

W�1��

� 2�17��Y1Y

W2

2

W�3��

L3 � �∂∂W

C

3

� � �3Y1 � �6Y2 � 2�9�Y1Y2� � �11Y1��W

W1

3

�� 12Y1��W

W2

3

�� 14Y2��

W

W2

3

�� 15Y2��W

W3

1�� 2�17��

Y1Y

W2

3

W�2��

� 2�18��Y1

W

Y2

3

W�1��,

where L1, L2, and L3 denote managerial, technical, and blue-collar work-ers, respectively.

We may use seemingly unrelated regressions to estimate equation (1),taking into consideration the fact that disturbance terms in the three singleequations may be somehow correlated. In undertaking the regression, weshould impose cross-equation restrictions on parameters to ensure that thesame estimate is produced for any parameter that appears in more than oneequation. From the parameter estimates, we can easily measure the effectsof Y1 and Y2 on each kind of labor demand, as shown in equation (1).

In order to measure the quantity of labor, data was drawn from the lat-est survey on employment undertaken by Taiwan’s Bureau of Labor Af-fairs (BOLA) in 1999. The survey classifies labor into nine categories, butthese nine categories are far too many to handle and also contain manyzeros; therefore, they are combined into three categories to suit our pur-poses: (a) supervisory (managers), administrative, and professional staffare classified as managerial workers; (b) engineers, technicians, and spe-cialists are classified as technical workers; and (c) operators, laborers, andservice workers are classified as blue-collar workers. The raw data drawn


from the three small labor categories are converted into a large category,using the Divisia index, with each sample mean being normalized to unity.We thus obtained the measures for L1 (managerial workers), L2 (technicalworkers), and L3 (blue-collar workers).

Wage rates W1, W2, W3 are obtained by dividing the respective total wagebills by the measures of L1, L2, and L3. The data for domestic output (Y1) andoverseas output (Y2) are obtained from the 1999 Survey on Overseas Invest-ment by Manufacturing Firms undertaken by the Ministry of EconomicAffairs (MOEA). This survey also provides information on investment loca-tions, but it only covers manufacturing firms that possess overseas affiliates.The BOLA and MOEA surveys are combined to yield 394 observations, allof which are firms engaged in FDI. We then randomly drew 140 non-FDIfirms from the BOLA survey in order to supplement the observations usingfirms without overseas affiliates. The total of 140 was taken so as to make theratio of FDI to non-FDI firms roughly 3:1. The combined sample of 534firms form the basis of our regression analysis, but only 451 of them containcomplete data for entry into the regression estimation. It is generally be-lieved that Taiwanese firms underreported their actual amounts of invest-ment in China. Our usage of output value rather than investment amount inregression analysis avoids the underestimation problem. There may also befirms that hide their investment altogether. Hopefully, our randomly chosennon-FDI sample does not contain many of such firms. Both the MOEA andthe BOLA surveys covered firms of all sizes, so there is no selection biasproblem associated with size. The regression results are shown in table 4.5.

From equation (1), we can derive the effects of domestic output (Y1 ) andoverseas output (Y2 ) on labor demand. They are, respectively,

(2) �∂∂L

Y1

1� � �1 � �7��

Y

Y2

1

�� 10��W

W2

1

�� 11��W

W3

1

�� 16��Y

Y2

1

W

W2

1

�� 18��Y

Y2

1

W

W3

1

��∂∂L

Y1

2� � �2 � �8��

Y

Y2

1

�� 10��W

W1

2

�� 12��W

W3

2

�� 16��Y

Y2

1

W

W1

2

�� 17��Y

Y2

1

W

W3

2

��∂∂L

Y1

3� � �3 � �9��

Y

Y2

1

�� 11��W

W1

3

�� 12��W

W3

2�� 17��

Y

Y2

1

W

W2

3

�� 18��Y

Y2

1

W

W1

3

��and

(3) �∂∂L

Y2

1� � �4 � �7��

Y

Y1

2

�� 13��W

W2

1

�� 15��W

W3

1

�� 16��Y

Y1

2

W

W2

1

�� 18��Y

Y1

2

W

W3

1

��∂∂L

Y2

2� � �5 � �8��

Y

Y1

2

�� 13��W

W1

2

�� 14��W

W3

2

�� 16��Y

Y1

2

W

W1

2

�� 17��Y

Y1

2

W

W3

2

��∂∂L

Y2

3��6 � �9��

Y

Y1

2

�� 14��W

W2

3

�� 15��W

W1

3

�� 17��Y

Y1

2

W

W2

3

�� 18��Y

Y

2

1

W

W1

3

��.


If we fit the parameter estimates into equations (2) and (3), we obtain theestimated effects of Y1 and Y2 on labor demand. The values of Y1 and Y2 ,and W1, W2 , and W3 , are taken to be the sample means. We estimate theseeffects for firms investing in different locations as we did in the previoussection. The results are shown in table 4.6.

It can be seen from table 4.6 that the demand for all kinds of labor in-creases with an increase in domestic output. For example, for those firmsinvesting in China only, the demand for managerial workers increases by0.1760 for each NT$ billion (Taiwanese currency) increase in domesticoutput (as Y1 is measured in NT$ billions). Since the Divisia index for la-bor has been normalized, this figure implies that in comparison with the


Table 4.5 Regression estimates of generalized Leontief production function

Independent variables Parameter estimates t-statistic

Dependent variable: Managerial workers (L1)Y1 –1.639 � 10–2 0.299Y2 0.391 3.110∗∗YY –0.232 2.430∗∗Y1W12 9.552 � 10–2 1.348Y1W13 0.103 3.152∗∗Y2W12 –0.449 3.068∗∗Y2W13 –3.610 � 10–2 0.992YYW12 0.363 3.016∗∗YYW13 –5.645 � 10–2 1.187

Dependent variable: Technical workers (L2)Y1 0.389 3.014∗∗Y2 0.739 3.985∗∗YY –0.748 3.997∗∗Y1W21 9.552 � 10–2 1.348Y1W23 –0.131 2.230∗∗Y2W21 –0.449 3.068∗∗Y2W23 –0.125 2.101∗∗YYW21 0.363 3.016∗∗YYW23 0.177 2.254∗∗

Dependent variable: Blue-collar workers (L3)Y1 0.205 4.116∗∗Y2 0.195 3.514∗∗YY –0.160 –2.541∗∗Y1W31 0.103 3.408∗∗Y1W32 –0.131 –2.380∗∗Y2W31 –3.610 � 10–2 0.992Y2W32 –0.125 2.101∗∗YYW31 –5.645 � 10–2 1.187YYW32 0.177 2.254∗∗

Notes: System weighted R2 � 0.5649; degree of freedom: 1,335; YY � (Y1Y2 )1/2; Y1W12 �Y1W1

–1/2 W21/2; Y1W13 � Y1W1

–1/2; Y2W12 � Y2W1–1/2 W2

1/2; Y2W13 � Y2W1–1/2 W3

1/2; YYW12 �(Y1Y2 )1/2W1

1/2; YYW13 � (Y1Y2 )1/2W1–1/2W3

1/2.∗∗Significant at the 5 percent level.

sample mean, there is an increase of 17.60 percent in managerial workers.Similarly, for each NT$ billion increase in domestic output, the demand fortechnical workers increases by 29.88 percent, and the demand for blue-collar workers increases by 14.12 percent. The results indicate that by 1999,the expansion in domestic production had led to an expansion in all threekinds of labor, although technical personnel tended to benefit the most,followed by managerial staff, and then blue-collar workers the least. Thispattern prevails across all investment locations, despite the fact that firmsize differs significantly across different subgroups. This implies that theoutput effect on employment is mainly driven by the nature of technologythat, as Taiwanese industry intensifies its technology content, tends to fa-vor technical workers.

Table 4.7 lists the mean values of Y1 and Y2 for the different FDI sub-groups. It can be seen that the subgroup of firms investing in China only isthe smallest of the three groups in terms of domestic output, followed bythe subgroup investing in China plus other regions, with the subgroup in-vesting only outside of China being the largest. However, the subgroup in-vesting in China and other regions also has the highest overseas productionratio, at 0.702, followed by the China only subgroup at 0.475, and then theoutside China subgroup at 0.292.

Referring back to table 4.6 also shows that overseas production has ex-erted a uniformly negative effect on each kind of labor, which suggests that


Table 4.6 Effects of domestic and overseas production on employment

Managerial Technical Blue-collar

Domestic Overseas Domestic Overseas Domestic Overseas production production production production production production

Investing in China only (136) 0.1760 –0.0291 0.2988 –0.0413 0.1412 –0.0481

Investing in China and others (126) 0.1847 –0.0286 0.2831 –0.0387 0.1264 –0.0220

Investing outside China (113) 0.1762 –0.0307 0.3018 –0.0533 0.1559 –0.0845

Notes: Domestic and overseas production is estimated in NT$ billions. Number of samples in paren-theses.

Table 4.7 Sample means, by FDI group (NT$ million)

Domestic Overseas Overseas/Domestic No. of FDI location output output ratio samples

China only 1,795.3 851.9 0.475 136China and others 3,995.0 2,805.8 0.702 126Other than China 5,591.1 1,633.8 0.292 113

when holding domestic output constant, domestic employment for a firmengaging in overseas production will decline by between 2 percent and 8percent. This implies that overseas production complements domesticproduction and therefore reduces the need for labor inputs at any givenoutput level. However, we should not jump to the conclusion that overseasproduction reduces domestic employment, because such a complementaryrelationship also cuts down the cost of domestic production, thus enhanc-ing the competitiveness of the company as a whole, which, in turn, maylead to an expansion in domestic output. In other words, overseas produc-tion exerts a substitution effect that reduces the demand for labor at anygiven domestic output as well as an output effect that expands domesticproduction. The net result has to take both effects into account; thus, it isthe output effect to which we now turn.

We take the Census of Manufacturers data and choose the firms that havesurvived throughout the period under study to explore the effects of FDIon domestic output. A simple regression is employed to estimate this effect:

(4) LY99 � �0 � �1LY93 � �2DFI1 � �3DFI2 � �4DFI3 � �5DFI4

� �6IND

where the variables are as follows:

LY99: logarithm of domestic output in 1999LY93: logarithm of domestic output in 1993DFI1: dummy variable for firms investing in China onlyDFI2: dummy variable for firms investing in China and other regionsDFI3: dummy variable for firms investing only outside ChinaDFI4: dummy variable for firms investing in unknown regionsIND: dummy variable for high-growth industries

In equation (4), we use the output in the base year (i.e., 1993) to projectthe output in the future year, 1999. Thus the coefficient �1 reflects the av-erage growth rate between 1993 and 1999. The dummy variables, DFI1–DFI4, capture the extra growth attributable to overseas investments, andthe dummy variable, IND, captures the extra growth attributable to indus-try affiliation. Included in the regression analysis were a total of 50,164firms that survived the 1993–1999 period. The results are reported in table4.8, which shows that the coefficients for dummy variables, DFI1–DFI4,were all positive and statistically significant. This suggests that foreign in-vestment does indeed contribute to extra growth in output after control-ling for the industry effect.

Compared to non-FDI firms, firms investing only in China recorded ex-tra growth of 18 percent over the 1993–1999 period, those firms that in-vested in China and other regions gained an extra 51.7 percent, and thosewhose investment was only outside of China achieved 46.4 percent growth.


The gains may be different, but other things being equal, FDI has indeedexpanded their domestic output.

We can therefore estimate the output effect of FDI on domestic produc-tion using these estimates; that is, our aim is to estimate the additional do-mestic output that is attributable to FDI.

Taking the estimate of � in equation (4), this would be �Y1 � Y1 � �/(1 ��), where � corresponds to the location of investment. This output effect isto be added to the substitution effect to come up with the net effect of over-seas production on domestic labor demand; thus, the total effect of FDI ondomestic labor Li is

(5) �Li � �∂∂Y

L

1

i� �Y1 � �

∂∂Y

L

2

i��Y2 ,

where the first term reflects the output effect and the second term reflectsthe substitution effect.

Inserting the relevant parameter estimates into equation (5), using therelations established in equation (4), we obtain the estimates at the samplemeans for total employment effect arising from FDI. These are shown intable 4.9.

It can be seen from table 4.9 that the total employment effects on FDIare positive for all kinds of labor and for all investment locations, withthe exception of those investments undertaken outside of China. For thesubgroup investing only outside of China, domestic employment of blue-


Table 4.8 Effect of FDI on domestic output

Dependent variable: LY99 Parameter estimates t-statistic

Intercept 1.217 44.562∗∗LY93 0.869 303.763∗∗Investing in China only (DFI1) 0.180 5.573∗∗Investing in China and others (DFI2) 0.517 9.288∗∗Investing outside China (DFI3) 0.464 13.071∗∗Unknown FDI regions (DFI4) 0.424 10.530∗∗High-growth industry (IND) 0.198 21.388∗∗

Notes: R2 � 0.6818; F-statistic � 17,915.45; degrees of freedom: 50,158.∗∗Significant at the 5 percent level.

Table 4.9 Overall effect of FDI on domestic employment

Managerial Technical Blue-collar workers workers workers

Investing in China only 0.0402 0.0717 0.0277Investing in China and others 0.1185 0.1933 0.0755Investing outside of China 0.0833 0.1415 –0.0237

collar workers is adversely affected by FDI (a decline of 2.37 percent). Thetable also shows that technical workers are the biggest winners from FDI;regardless of the investment locations, the greatest increase is in the do-mestic employment of technical workers. We interpret this outcome as re-flecting the fact that domestic production in recent years has been restruc-tured towards more technology-intensive methods. Managerial workersalso gain substantially from FDI but not as much as their technical coun-terparts. Blue-collar workers gain the least, and they may occasionallyeven lose. Capital outflow favoring technical workers was also found inFeenstra (1996), while Blomström, Fors, and Lipsey (1997) found that it fa-vored managerial staff. In short, FDI may well affect different labor groupsin different ways, but the overall effect is more likely to be positive thannegative. The group that is most likely to feel any negative effects is theblue-collar group of workers.

It is noticeable that firms simultaneously investing in China and otherregions create the greatest proportion of new jobs at home. We take thissubgroup of firms to be truly in pursuit of globalization, since globaliza-tion leads to an expansion of domestic production. This also manifestsitself in the largest parameter estimate for DFI2 among all DFIs. Thoseinvesting only in China do not create as much demand for technical andmanagerial workers at home because production in China is characterizedby a low technology requirement and simple production arrangements.

Going back to table 4.3 in which domestic employment is shown to de-cline for firms investing only in China, we may conclude that FDI, per se,is not to blame for the plight of labor; it is instead the fact that these in-vestors belong to low-growth (or even declining) industries, as well as be-ing small in size, that account for their inability to maintain their employ-ment levels at home. In addition to the industry effect, the fact that theChina-only group did not generate as much output-expansion effect as theother investment groups also contributes to their below-par performance.Although China production enhances the competitiveness of domesticproduction, just like other overseas production, it also takes market op-portunities away from Taiwan because Chinese and Taiwanese suppliersare often viewed by foreign buyers (particularly in the Western markets) asclose substitutes.

4.6 Conclusions

In this paper we study the effects of FDI on domestic employment by ex-amining the data of Taiwan’s manufacturing industry. In terms of growthin their number of employees, those firms investing abroad have outper-formed those firms that have not undertaken such investment. Moreover,firms that have invested abroad have a higher probability of survival thanthe have nots; survival means maintaining some jobs at home.


Treating domestic production and overseas production as two distinctbut interrelated outputs from a joint production function, we may estimatethe effects of overseas production on domestic production and, thereafter,the consequences for domestic employment. Our study of Taiwanese man-ufacturing data indicates that overseas production reduces the demand forlabor in domestic operations at any given domestic output. This impliesthat through joint production, overseas production reduces the input re-quirements at home to yield a given domestic output. In other words, over-seas production substitutes for primary inputs in the domestic productionprocess.

From a presumption of cost-minimization, this implies that overseasproduction complements domestic production to reduce the overall costsof cross-border operations, thereby enhancing the competitiveness of acompany; this is to be achieved through a division of labor between theheadquarters and the affiliates. Such enhanced competitiveness, in turn,helps firms to expand their domestic output, which leads to an increasein the demand for labor. Therefore, the total effect of FDI on domesticemployment is a combination of output-expansion effect and input-substitution effect. Our estimates show that, in most cases, the output-expansion effect more than offsets the input-substitution effect to yield anet positive effect on domestic employment; however, the magnitude ofemployment effect arising from FDI differs across different labor groups.

In the case of Taiwan, technical workers tend to benefit most from FDI,followed by managerial workers, with blue-collar workers benefiting theleast; indeed, they may even be adversely affected. This implies that afteroverseas investment has taken place, a reconfiguration of the division of la-bor within a firm will tend to shift domestic production toward technol-ogy- and management-intensive operations.

Different investment locations exert slightly different impacts on do-mestic employment mainly because of the differences in output-expansioneffect. Those firms that invest only in China contribute the least to the ex-pansion of domestic output, followed by firms that invest only outside ofChina, while FDI covering both China and other regions is most con-ducive to domestic output expansion.

References

Aw, B.-Y., X. Chen, and M. Roberts. 2001. Firm-level evidence of productivity dif-ferentials and turnover in Taiwanese manufacturing. Journal of DevelopmentEconomics 66 (1): 51–86.

Bayoumi, T., and G. Lipworth. 1998. Japanese foreign direct investment and re-gional trade. Journal of Asian Economics 9 (4): 581–607.


Blomström, M., G. Fors, and R. Lipsey. 1997. Foreign direct investment and em-ployment: Home country experience in the United States and Sweden. EconomicJournal 107 (445): 1787–97.

Blomström, M., and A. Kokko. 2000. Outward investment, employment and wagesin Swedish multinationals. Oxford Review of Economic Policy 16 (3): 76–89.

Brainard, S., and D. Riker. 1997a. Are US multinationals exporting US jobs?NBER Working Paper no. 5958. Cambridge, MA: National Bureau of Eco-nomic Research, March.

———. 1997b. US multinationals and competition from low-wage countries.NBER Working Paper no. 5959. Cambridge, MA: National Bureau of Eco-nomic Research, March.

Caves, R. 1971. International corporations: The industrial economics of foreigninvestment. Economica 38:1–27.

———. 1996. Multinational enterprises and economic analysis. Cambridge, MA:Cambridge University Press.

Chen, T. J., and Y.-H. Ku. 2000. Foreign direct investment and industrial restruc-turing: The case of Taiwan’s textile industry. In The role of foreign direct invest-ment in East Asian economic development, ed. I. Takatoshi and A. Krueger, 319–48. Chicago: University of Chicago Press.

Diewert, W. E. 1971. An application of the Shephard duality theorem: A general-ized Leontief production function. Journal of Political Economy 79 (3): 481–507.

Döhrn, R. 1997. Bestimmungsgründe von umfang und entwicklung der Ausland-saktivitäten deutscher unternehmen [Determinants of the scope and develop-ment of overseas activities of German enterprises]. Lecture notes. Essen; RWIInstitute.

Feenstra, R. 1996. Foreign investment, outsourcing and relative wages. In Politicaleconomy of trade policy: Essays in honor of Jagdish Bhagwati, ed. R. Feenstra andG. Douglas, 89–127. Cambridge, MA: MIT Press.

Feldstein, M. 1994. The effects of outbound foreign direct investment on the do-mestic capital stock. NBER Working Paper no. 4668. Cambridge, MA: NationalBureau of Economic Research, March.

Hall, R. 1973. The specification of technology with several kinds of output. Jour-nal of Political Economy 81 (4): 878–92.

Hatzius, J. 1997. Domestic jobs and foreign wages: Labour demand in Swedishmultinationals. CEP Discussion Paper no. 337. London: Centre for EconomicPerformance.

Ku, Y.-H. 1998. Foreign direct investment and domestic restructuring: The case ofTaiwan’s electronic industry. Taiwan Economic Review 26 (4): 459–86.

Lipsey, R. 1994. Outward direct investment and the US economy. NBER Work-ing Paper no. 4691. Cambridge, MA: National Bureau of Economic Research,March.

Liu, B.-J., and H.-L. Lin. 2001. Reverse imports and outward investment. TaiwanEconomic Review 29 (4): 479–510.

Markusen, J. 1983. Factor movements and commodity trade as complements. Jour-nal of International Economics 14 (3–4): 341–56.

Mundell, R. 1957. International trade and factor mobility. American Economic Re-view 47:321–35.

Slaughter, M. 1995. Multinational corporations, outsourcing and American wagediversion. NBER Working Paper no. 5253. Cambridge, MA: National Bureau ofEconomic Research, September.

Stevens, V., and R. Lipsey. 1992. Interactions between domestic and foreign invest-ment. Journal of International Money and Finance 11 (1): 40–62.


Comment Keiko Ito

This paper is very well written and organized and presents very interestingfindings. It conducts rigorous quantitative analyses based on a couple oflarge sets of microdata and investigates the relationship between outwardforeign direct investment (FDI) and domestic employment at the firmlevel. The authors argue that overseas production substitutes for the do-mestic production of primary inputs, which reduces the demand for laborin domestic operations; on the other hand, overseas production also low-ers the costs of cross-border operations, which enhances companies’ over-all competitiveness and, in turn, helps them to expand domestic output.On the whole, therefore, FDI has a net positive effect on domestic employ-ment. Because there are very few studies on this issue, particularly at sucha microlevel, for middle-developed countries like Taiwan, this is a highlycommendable paper.

The study makes an important contribution in the following respects.First, it deals with a very interesting and important topic. The effect of out-ward FDI on domestic labor demand is an issue of major concern in manydeveloped countries, where people are worried about job losses as manyproduction processes are shifted to low-wage countries. Although a largenumber of studies have been conducted for the United States and Euro-pean countries, very few have looked at Asian countries to examine thistopic. The second contribution of this paper is that it analyzes firm-leveldata. The authors use three sets of microdata—from the ManufacturingCensus, the employment survey by Taiwan’s Bureau of Labor Affairs, andthe Survey on Overseas Investment by Manufacturing Firms by the Min-istry of Economic Affairs—and combine the latter two data sets for theiranalysis. Third, the authors compile and analyze simple statistics and alsoconduct an econometric analysis of the data based on the generalizedLeontief cost function framework. Both the simple analysis of the statisticsin itself and their econometric investigation, which makes it possible to es-timate the various effects of FDI on domestic labor demand quantitatively,provide valuable insights. Although we should note that their econometricanalysis may include substantial errors and biases, it is certainly the lastpoint that is the most important contribution of this paper. I will discussthe details of the econometric method later.

First, however, I would like to turn to specific comments. Table 4.1 showsthat almost half of Taiwan’s total outward FDI goes to China, based onwhich the authors focus on China as the main destination for FDI in theanalysis that follows. In the following analyses, they divide sample firms ac-cording to the location of their FDI into four categories: China only, China


Keiko Ito is a lecturer of economics at Senshu University.

and others, other than China, and unknown. We should note, however,that, according to table 4.1, the United States is another large recipient ofTaiwanese FDI, also accounting for almost a half of Taiwan’s total FDI.Taiwan’s FDI in the United States is conducted probably for the purposeof establishing a sales base, while FDI in China is for production. Thisdifference may have a crucial influence on the estimated effects of FDI onthe domestic demand for labor. In addition, the fact that the United Statesis such a large recipient of Taiwanese FDI implies that many firms in the“China and others” category would have undertaken investment in Chinaand the United States. These firms are likely to produce in China and selltheir products in the United States. The results throughout the paper mightsuggest that firms with both production and sales bases overseas (i.e., FDIcovering both China and other regions) are most likely to experience an ex-pansion of domestic output and employment.

Next, table 4.2 suggests that firms engaging in FDI have a higher sur-vival rate than those that do not. We should point out that non-FDI firmsare much smaller than FDI firms. Because larger firms are more likely tosurvive, the higher survival rate of FDI firms might be attributable to thedifference in size of non-FDI firms and FDI-firms. In addition, I am notsure how the authors treat firms that were not undertaking FDI as of 1993but were doing so as of 2000. These firms should have been in the “non-FDI firms” cohort in 1993 and entered the “FDI firms” cohort in 2000. Inthat case, are they treated as exited non-FDI firms and as new entrantsamong FDI firms? If the authors treat these firms like this, the exit rate fornon-FDI firms may be exaggerated. It should be made clear whether non-FDI firms may only seem to have exited business, while in fact they onlymoved from one category to another.

Finally I would like to comment on the econometric analysis employedin this study. This study estimates labor demand functions derived from thegeneralized Leontief cost function, while most related previous studies em-ploy the translog cost function approach. Probably because capital stockdata at the firm level are not available, the authors assume that the elastic-ity of substitution between capital and composite labor is zero and there-fore estimate the generalized Leontief function without capital input.However, I think that the strong assumption on the substitution betweencapital and labor may induce serious errors and derive biased estimatesof cost parameters between subgroups in the labor force. According toHamermesh’s (1986) literature survey, previous empirical investigationsconclude that the separability of labor from capital is not supported by thedata and suggest that it is necessary to include the quantity or price of cap-ital services. We should also note that the estimation may include furthererrors or biases due to the exclusion of factor inputs for overseas produc-tion. Given the unavailability of capital stock (or capital price) data andoverseas factor inputs (or factor price) data at the firm level, there might be


no choice but to exclude these inputs. However, we have to be careful in in-terpreting the estimated effects of domestic and overseas production onemployment. The most consistent finding in previous studies on labor de-mand is that physical capital substitutes more easily for production work-ers (unskilled labor) than for nonproduction workers (skilled labor;Hamermesh 1986). Moreover, most previous studies found that the de-mand elasticity for nonproduction workers is lower than that for produc-tion workers (Hamermesh 1986). Given the decreasing trend in the priceof capital in Taiwan, we may infer that the demand for blue-collar workerswill decrease much more than the demand for managerial or technicalworkers when the capital input data are included in the analysis. Further-more, if the wage level of blue-collar workers in Taiwan rose above that inChina, the domestic demand for blue-collar workers among firms invest-ing in China would be further reduced. It is certainly very difficult to in-clude the price of overseas factor inputs due to data constraints. However,even when capital stock data are not available, it might still be possible tocalculate or estimate the price of capital at the firm level, for example, byobtaining data on interest payments, the amount of debt, and so on, ifthese are available. Alternatively, the authors could estimate the price ofcapital at the industry level and use this for their cost function analysis, as-suming that the price of capital is the same for all firms within an industry.

However, these are minor criticisms and suggestions for further expan-sion of a study that, overall, is solid and shows very interesting results. Theauthors were able to show that overseas production in most cases had a netpositive effect on domestic labor demand and that technical workers werethe biggest winners from FDI, while blue-collar workers gained the least.This implies that domestic production in Taiwan is tending to shift towardtechnology- and management-intensive operations. The observation thatthe shift in skilled-labor occurs in a middle-developed country like Taiwanwill attract a lot of attention. However, we should be aware of possible er-rors or biases as a result of the exclusion of capital and overseas factor in-puts, and I hope that the authors will get a chance to include these factorinputs in future studies.

Reference

Hamermesh, Daniel S. 1986. The demand for labor in the long run. In Handbookof labor economics. Vol. 1, ed. O. Ashenfelter and R. Layard, 429–71. Amster-dam: Elsevier Science.


Comment Francis T. Lui

The paper by Tain-Jy Chen and Ying-Hua Ku raises an interesting empir-ical question that does not have an unambiguous theoretical answer. Doesforeign direct investment (FDI) made by the home country cause job lossesat home? Theoretically, there are two opposite forces at work. Overseasproduction substitutes exports. Reduction in exports lowers employmentat home. On the other hand, a firm that makes overseas investment tendsto be better able to use resources efficiently. This helps the firm to survivethe competition at home and enables it to continue to hire workers. If theefficiency gain is big enough, the firm may even expand itself and increaseemployment.

The question of which force is stronger has important policy implica-tions. Labor unions may believe that FDI hurts workers at home, andtherefore they lobby against it. This question is also more general than itappears to be. Analytically it is similar to another policy question. Coun-tries that contemplate importing labor from outside may face the same op-position. A firm using its money to hire imported workers in the homecountry is in a similar situation compared to another that sends money toother countries in the form of FDI. Importing labor also has two oppositeeffects. First, some local workers are displaced. Second, importing cheaperor more-efficient workers makes the firm more competitive at home. Thechance for its migration to other countries decreases. Thus, there is also ajob creation effect. The answer to whether labor should be imported isagain an empirical issue.

The paper has set an example for policy research that is based on solidempirical findings and rigorous methodology. It assumes that the produc-tion function is generalized Leontief. By making use of the Shephard’slemma and estimating the cost function, the paper shows that a firm’s em-ployment of different types of workers is a function of their respective wagerates, its output at home and at the foreign country. It is found that an in-crease in the firm’s output at the foreign country reduces employment of alltypes of workers at home, provided that output at home is held constant. Itis also shown that overseas investment raises output at home, which in turnincreases employment. According to the paper’s estimates, the net effect ofoverseas investment on home employment is positive. The methodology isnicely developed, and the results are reasonable.

If overseas investment actually raises home employment, why do laborunions object to it? A plausible answer is that competition from abroad


Francis T. Lui is a professor of economics and director of the Center for Economic Devel-opment at the Hong Kong University of Science and Technology.

could lower the wage rates of workers at home. It would be useful if the au-thors could estimate the effect on wage rates.

The paper documents that there has been substantial restructuring ofthe Taiwanese economy during the sample period. From 1991 to 2000, theelectronics industry has employed 145,748 more people, but the apparel in-dustry has lost 54,104 jobs. Moreover, manufacturing has 400,000 fewerjobs, but the service sector has gained 1.4 million jobs in the period from1987 to 1994. When economic restructuring takes place, the unemploy-ment rate usually rises because people leaving an industry need time togenerate offers from other industries. One would suspect that economic re-structuring should be one of the factors that affect unemployment duringthe sample period. In a study on the effect of imported workers on em-ployment in Hong Kong, we have found that it is essential to include re-structuring as a control variable (Kwan, Lian, and Lui 1995). The latter canconveniently be measured by an index proposed by Lilien (1982), whichtells us how extensive are the movements of workers from one sector toanother. Our results indicate that economic restructuring significantlychanges the unemployment rate, but the number of imported workers haslittle or no effect at all. To reduce the possibility of specification bias dueto the omission of an important variable, the authors may want to addressthis issue in their future research.

Because globalization and fragmentation will likely continue, overseasinvestment will grow. The findings of this paper will become more andmore relevant and important in the future.

References

Kwan, Yum-Keung, Y. Joseph Lian, and Francis T. Lui. 1995. Hong Kong’s unem-ployment and imported labour. Hong Kong University of Science & Technology.Mimeograph.

Lilien, David M. 1982. Sectoral shifts and cyclical unemployment. Journal of Po-litical Economy 90 (4): 777–93.


the effect of overseas investment on domestic employment · the majority of overseas investment,...

Documents